Inhibition Mechanisms of (-)-Epigallocatechin-3-gallate and Genistein on Amyloid-beta 42 Peptide of Alzheimer's Disease via Molecular Simulations.

The misfolding and self-assembly of amyloid-beta (Aβ) peptides are one of the most important factors contributing to Alzheimer's disease (AD). This study aims to reveal the inhibition mechanisms of (-)-epigallocatechin-3-gallate (EGCG) and genistein on the conformational changes of Aβ42 peptides by using molecular docking and molecular dynamics (MD) simulation. The results indicate that both EGCG and genistein have inhibitory effects on the conformational transition of Aβ42 peptide. EGCG and genistein reduce the ratio of β-sheet secondary structures of Aβ42 peptide while inducing random coil structures. In terms of hydrophobic interactions in the central hydrophobic core of Aβ42 peptide, the binding affinities of EGCG are significantly larger in comparison with that of genistein. Our findings illustrate the inhibition mechanisms of EGCG and genistein on the Aβ42 peptides and prove that EGCG is a very promising inhibitor in impeding the conformational change of Aβ42 peptide.

Introduction

Alzheimer’s disease (AD), which can cause cognitive functional disorder and behavioral impairment, severely affects the quality of life of the elderly. In most cases, AD has two main pathological features, which are extraneuronal plaques of misfolded amyloid-beta (Aβ) proteins and intraneuronal neurofibrillary tangles of hyperphosphorylated tau protein in the brain.[1] Especially, the extracellular Aβ plaques produced by Aβ peptides’ aggregation are one of the unique characteristics of AD.[2] The two main components of Aβ peptides are 40 amino acid residues (Aβ40 for short) and 42 amino acid residues (Aβ42 for short). In fact, monomeric Aβ42 peptide plays a key role in the initial development of Aβ plaques.[1] Oligomers are formed by the self-assembly of misfolded Aβ42 peptides and then further aggregate to form fibrils. Then gradually, plaques deposit around nerve cells in the brain (Figure ).

Figure 1

Schematic illustration of possible mechanisms of inhibiting the conformational change of Aβ42 peptide by EGCG and genistein.

Aβ42 monomer is a very important marker of AD, which has been widely used in the prevention and treatment of AD.[3] The pathway to prevent the misfolding and further aggregation of monomeric Aβ42 peptide is either to inhibit conformational transition of Aβ monomer by targeting inhibitors or to stabilize the native state of Aβ42 monomer by refolding of the misfolded conformation.[4] As to conformational transition inhibitors of Aβ42 monomer, natural polyphenolic compounds extracted from edible plants attract much attention because of their low toxicity and few side effects on the human body.[5]

(−)-Epigallocatechin-3-gallate (EGCG), the major polyphenolic constituent of green tea, has been proven to have a neuroprotective effect on aging and neurodegenerative diseases.[6,7] Harvey et al. have demonstrated that EGCG is able to disaggregate the preformed Aβ fibrils and protect neuronal cells in vitro.[8] Similarly, the green tea polyphenol EGCG can redirect the amyloidogenic aggregation pathway by expanding ataxin-3 toward nontoxic, soluble, SDS-resistant aggregates.[9] Acharya et al. have investigated the molecular mechanisms between EGCG and Aβ polypeptides by combining in vitro immuno-infrared sensor measurements, molecular docking, and molecular dynamics (MD) simulations. They show that the intermolecular interactions of EGCG and Aβ polypeptides are dominated by a few residues in the fibrils.[10] In addition, the interaction mechanisms of EGCG in inhibition of Aβ42 oligomers[11] and disaggregation of Aβ42 fibrils[12,13] have also been studied by using molecular simulations. The results illustrate that EGCG can prevent the aggregation of Aβ42 oligomers and increase the destabilization effect of Aβ42 fibrils.

On the other hand, soybean isoflavone is a typical polyphenol present extensively in soy foods, which has potential therapeutic effects on AD. Ding et al. have found that soybean isoflavones can alleviate the learning and memory deficit induced by Aβ42 peptide in rats by protecting the synapse structure and function.[14] Isoflavones are present in significant quantities in soybeans, which are called genistein (5,7,4′-trihydroxyisoflavone), daidzein (7,4′-dihydroxyisoflavone), and glycitein (7,4′-dihydroxy-6-methoxyisoflavone).[15] Among them, genistein is the main isoflavone, which has the advantage of systemic nontoxicity.[16] Furthermore, genistein not only has an excellent neuroprotective effect but also is a multitarget inhibitor.[17] Ma et al. have demonstrated genistein’s neuroprotective effects against Aβ-induced neuroinflammation through regulating the Toll-like receptor 4/nuclear factor κB signaling pathway.[18] Further, Petry et al. have found that genistein can protect against Aβ-induced toxicity in SH-SY5Y cells by inhibiting Aβ-induced protein kinase B inactivation and tau hyperphosphorylation.[19] Ren et al. have revealed that genistein strongly inhibits Aβ42 monomer self-aggregation at the very beginning of the aggregation. Moreover, MD results showed that genistein prefers to bind the β-sheet groove of Aβ42 pentameric protofibril.[20]

However, the mechanisms of how the EGCG molecule and genistein molecule inhibit the conformational transitions of Aβ42 monomers have not been well studied at present. In this work, we use molecular simulations to investigate the inhibiting behaviors of EGCG and genistein on the conformational changes of full-length Aβ42 peptides and explore the differences in molecular mechanisms (Figure ). Meanwhile, the specific binding patterns of EGCG and genistein with Aβ42 peptides are demonstrated by molecular docking simulation. MD simulation studies reveal structural stabilities, secondary structure distributions, and the effect of two polyphenolic molecules on the hydrogen bonds of Aβ42 peptides. In addition, the binding free energy and energy contribution per amino acid residue of Aβ42 peptide in systems with EGCG or genistein are calculated. In conclusion, both EGCG and genistein can inhibit the conformational change of Aβ42 peptide. The differences in the inhibitory mechanisms of EGCG and genistein are also discussed. Our studies provide a theoretical basis for the design of new drug candidates to inhibit the conformational transition and further self-assembly of Aβ42 monomer.

Results and Discussion

Binding Sites of Aβ42 Monomer

To investigate the specific binding sites of Aβ42 peptide, we adopt the DoGSiteScorer tool to acquire the prediction of binding regions. The binding pockets of Aβ42 monomer are predicted to be pocket 1 (Leu17, Phe20, Ala21, Gly25, Lys28, Ile31, Ile32, Leu34, and Met35) and pocket 2 (Ser8, Glu11, Val12, Gln15, and Lys16), which are shown in Figure . The interaction mechanisms between the two specific binding regions of Aβ42 peptide and two polyphenolic molecules are further examined by molecular docking simulation.

Figure 2

Predicted binding sites of Aβ42 peptide: pocket 1 is shown in gray mesh, and pocket 2 is shown in pink mesh.

Molecular Docking Simulation of Aβ42 Peptide with Respect to EGCG and Genistein

Autodock can offer information on the interactions of two polyphenolic molecules with the two binding pockets of Aβ42 peptide. For interactions between pocket 1 of Aβ42 peptide and EGCG, the binding energies are −3.33 kcal/mol and −2.69 kcal/mol for the first and second poses of EGCG, respectively. And the binding energies are −4.28 kcal/mol and −2.85 kcal/mol for the interactions of pocket 2 of Aβ42 peptide with the first and second conformations of EGCG. However, only one docking pose of genistein is proposed for the interactions between genistein and pocket 1 of Aβ42 peptide. The binding energy is −4.24 kcal/mol. Similarly, only one docking conformation is also obtained between pocket 2 of Aβ42 peptide and genistein. The binding energy is −5.10 kcal/mol. Since the negative sign means favorable binding pose,[21] the lowest values for EGCG (−4.28 kcal/mol) and genistein (−5.10 kcal/mol) illustrate that the conformations of them strongly bind to binding site 2 of Aβ42 peptide. Therefore, they are used as optimal candidates. The intermolecular interactions of binding region 2 of Aβ42 peptide with the favorable binding conformations of small molecules are considered for further investigations (Figure ).

Figure 3

Intermolecular interactions between binding region 2 of Aβ42 peptide and the optimal binding conformations of EGCG and genistein. (a) Aβ42 peptide and EGCG; (b) binding region 2 of Aβ42 peptide and EGCG; (c) Aβ42 peptide and genistein; (d) binding region 2 of Aβ42 peptide and genistein. The hydrogen bonds are shown as purple dotted lines, while the marked residues of Aβ42 peptide make nonbonded contacts with the small molecules.

As shown in Figure a and b, the interactions between the favorable binding pose of EGCG and residues of Aβ42 peptide are depicted. The amine group of residue Gln15 is the hydrogen bond donor, and the ester and hydroxyl groups of EGCG are the hydrogen bond acceptors for the hydrogen bonds (dO···H = 2.09 Å, dO···N = 2.99 Å, ∠NHO = 147.8° and dO···H = 2.14 Å, dO···N = 3.05 Å, ∠NHO = 161.1°). The carboxyl group of residue Asp7 is the hydrogen bond acceptor, and the hydroxyl group of EGCG is the hydrogen bond donor in the hydrogen bond (dO···H = 2.20 Å, dO···O = 3.05 Å, ∠OHO = 146.2°). In addition, hydrophobic interactions between EGCG and residues Asp7, Ser8, Glu11, Val12, Gln15, Lys16, and Phe19 of Aβ42 peptide are formed.

Figure c and d show that interactions are generated between the optimal binding conformation of genistein and residues of Aβ42 peptide. The hydrogen bond (dO···H = 2.15 Å, dO···N = 3.04 Å, ∠NHO = 145.9°) is formed by the amine group of residue Gln15 as the hydrogen bond donor and the carbonyl group of genistein as the hydrogen bond acceptor, and hydrophobic interactions between genistein and residues Ser8, Glu11, Val12, Gln15, Lys16, and Phe19 of Aβ42 peptide are formed. Taken together, all of the formed hydrogen bonds are stable according to the literature.[22] Both EGCG and genistein have strong interactions with the hydrophobic segment[23] of Aβ42 peptide, implying that EGCG and genistein can retard the amyloidogenic potential of monomeric Aβ42 peptide.

Molecular Dynamics Simulation of Aβ42 Peptide with Respect to EGCG and Genistein

Validation of Simulation Data

To verify the MD ensembles of the Aβ42 peptide–EGCG system and the Aβ42 peptide–genistein system obtained after a 100 ns simulation, the SHIFTX2 program[24] has been applied to evaluate the NMR chemical shifts of Aβ42 peptides. The correlation coefficients are obtained by calculating the chemical shifts of atoms Cα and Cβ in the Aβ42 peptide of the final MD ensembles and the Aβ42 peptide used initially for simulation. The NMR chemical shifts of the initial Aβ42 peptide are labeled as δexp, which come from experiments of Aβ42 peptide alone carried out in the medium of aqueous solutions of fluorinated alcohols,[25] and the NMR chemical shifts of the final Aβ42 peptide are labeled as δsim, which come from one unique generated structure at 100 ns. The final structures of Aβ42 peptide at 100 ns differ substantially from the initial configurations, as displayed in Figure .

Figure 6

Evolution of configurations of Aβ42 peptide (a) in water, (b) with EGCG in water, and (c) with genistein in water. Water molecules are not shown for clarity.

As shown in Figure a and b, for the simulation system of Aβ42 peptide–EGCG, the correlation coefficients of the chemical shifts of atoms Cα and Cβ are 0.93 and 0.99, respectively. Figure c and d shows the correlation coefficients of the Aβ42 peptide–genistein system. Similarly, the R values of the chemical shifts of atoms Cα and Cβ are 0.93 and 0.99, respectively. Herein, the high correlation coefficients are consistent with the reports in the literature.[26,27] The results turn out that the MD ensembles of the systems are reliable with the presence of EGCG and genistein.

Figure 4

Correlation of the NMR chemical shifts for atoms Cα and Cβ between the Aβ42 peptide used initially for simulation and the Aβ42 peptide of the final MD trajectories. (a and b) Aβ42 peptide–EGCG system; (c and d) Aβ42 peptide–genistein system. The unit of the NMR chemical shift is ppm.

Structural Analysis

To examine the stability of the results of MD simulations, the values of the root-mean-square deviation (RMSD) and radius of gyration (Rg) are calculated and the representative trajectories are extracted. As shown in Figure a, the RMSD values of three systems achieve stabilities after 70 ns, and the simulation time scale of 100 ns is sufficient to establish stable interactions of the Aβ42 peptide with two natural polyphenols. When the systems reach the first metastable states around the starting structures, the RMSD values have stabilized at about 1.27 ± 0.09 nm (Aβ42 peptide), 1.23 ± 0.05 nm (Aβ42 peptide–EGCG), and 1.13 ± 0.08 nm (Aβ42 peptide–genistein). Therefore, the 70–100 ns interval is selected for data collection and further analysis.

Figure 5

(a) Backbone RMSD of Aβ42 peptide in water and Aβ42 peptide with respect to EGCG and genistein. (b) Total Rg of Aβ42 peptide in water and Aβ42 peptide with respect to EGCG and genistein.

Figure b shows that the values of Rg of the three systems fluctuate greatly within 40 ns and are stabilized at 0.98 ± 0.04 nm (Aβ42 peptide), 1.03 ± 0.05 nm (Aβ42 peptide–EGCG), and 1.04 ± 0.06 nm (Aβ42 peptide–genistein). The results show that the Aβ42 peptides of three systems have undergone great conformational changes in the initial stage of the simulations and then have begun to stabilize after 70 ns. In addition, compared with the Rg value of the control system, the Rg values of the systems (Aβ42 peptide–EGCG and Aβ42 peptide–genistein) are slightly higher, which indicates that the compactness of Aβ42 peptide is reduced and the conformational changes of Aβ42 peptide are affected by the presence of EGCG and genistein.

In addition, representative snapshots of the three systems at different time points have been also compared. As shown in Figure a, the configuration of Aβ42 peptide in the control system has changed a lot at the beginning of the simulation trajectory. After the configuration of Aβ42 peptide is stable, β-sheet secondary structures are obviously observed. For the systems Aβ42 peptide–EGCG (Figure b) and Aβ42 peptide–genistein (Figure c), the configurations of Aβ42 peptide begin to undergo great changes and stretch gradually. After reaching the first metastable states around the starting structures, the secondary structures of Aβ42 peptide influenced by EGCG and genistein are dominated by random coils and α-helical structures. More detailed investigations of the secondary structural changes are discussed in the Secondary Structure Analysis section. Herein, the changing trends of the morphologies of three Aβ42 monomers are shown to be in agreement with those of RMSD and Rg.

In order to further compare the Aβ42 peptide influenced by EGCG and genistein, the values of the root-mean-square fluctuation (RMSF) and solvent accessible surface area (SASA) are examined. The RMSF value indicates the changes of amino acid residues in the Aβ42 peptide chain. As shown in Figure a, the fluctuations of amino acid residues in the Aβ42 peptide–EGCG system and the Aβ42 peptide-genistein system are more significant than that in the Aβ42 peptide system. Overall, the central hydrophobic core (L17VFFA21)[23] of Aβ42 peptide shows dramatic changes under the influence of EGCG. Compared to EGCG, genistein has a slighter effect on the hydrophobic segment and exhibits a stronger impact on the C-terminus of Aβ42 peptide.

Figure 7

(a) RMSF values for Aβ42 peptide in water and Aβ42 peptide with respect to EGCG and genistein. (b) SASA values for Aβ42 peptide in water and Aβ42 peptide with respect to EGCG and genistein.

Figure b displays the results of the solvent accessibility of the Aβ42 peptide surfaces in different simulation systems. The SASA value is very important for the influence of the configuration of a protein in water.[28] It is obvious that the values of SASA have stabilized at about 33.27 ± 1.12 nm/S2/N (Aβ42 peptide), 35.16 ± 3.10 nm/S2/N (Aβ42 peptide–EGCG), and 34.37 ± 1.50 nm/S2/N (Aβ42 peptide–genistein). This result means that the contact areas between the surfaces of Aβ42 peptide and water are increased under the action of EGCG and genistein. The SASA value of the Aβ42 peptide–EGCG system is slightly increased in comparison with that of the Aβ42 peptide–genistein system, implying that Aβ42 peptide can be more exposed to the water molecules in the presence of EGCG.

In sum, by combining RMSD, Rg, RMSF, and SASA data, the results demonstrate that EGCG and genistein show the inhibition of conformational changes against Aβ42 peptides, while the inhibition effect of EGCG on the conformational transition of Aβ42 peptide is better than that of genistein. The reason for this could be that the hydrophobic interactions between the EGCG molecule and Aβ42 peptide are stronger. Our findings are also fully similar to those of another simulation study by Li et al., who suggest that EGCG interacts with Aβ42 monomers mainly through hydrophobic interactions and inhibits the formation of Aβ42 dimers.[12]

Secondary Structure Analysis

It has been reported that the conformational conversions of peptide from the initial α-helix to β-sheet are then reorganized into a more organized β-sheet-richer structure, which ultimately leads to Aβ amyloidogenesis.[29] To demonstrate how to inhibit the conformational changes of Aβ42 peptide by EGCG and genistein, the definition of secondary structure of proteins (DSSP) is employed to offer information on the secondary structure contents of Aβ42 peptide in three simulated systems (Table ). Figure b and c displays the time evolution of the secondary structures of Aβ42 peptide in the systems with EGCG and genistein, respectively. Because the balance of helical elements is not produced correctly by the parameter set of the GROMOS force field family,[30] the sum of helical structures is marked as the α-helix.[31]

Table 1

Comparison of Secondary Structure Components in the Systems Aβ42 Peptide, Aβ42 Peptide–EGCG, and Aβ42 Peptide–Genistein

secondary structure component (%)	coil	β-sheet	β-bridge	bend	turn	α-helixa
Aβ42 peptide	24.69 ± 4.58	2.75 ± 1.82	2.45 ± 0.86	14.91 ± 4.91	9.92 ± 6.06	45.28 ± 2.42
Aβ42 peptide–EGCG	28.00 ± 8.02	0	0.29 ± 0.05	15.40 ± 7.93	11.31 ± 6.94	45.01 ± 7.44
Aβ42 peptide–genistein	29.52 ± 8.90	0	0.12 ± 0.02	13.59 ± 6.89	13.01 ± 3.95	43.76 ± 12.85

The 5-helix and 3-helix in Figure correspond to the α-helix.

Figure 8

Time evolution of the secondary structures of Aβ42 peptide in (a) the Aβ42 peptide system, (b) the Aβ42 peptide–EGCG system, and (c) the Aβ42 peptide–genistein system. The vertical axis represents the residue numbers of Aβ42 peptide, and the horizontal axis represents the simulation time in nanoseconds. The secondary structures are color-coded.

As shown in Table and Figure , the proportion of α-helix is 45.28% for the system Aβ42 peptide. The average values of α-helical structures are 45.01% (Aβ42 peptide–EGCG) and 43.76% (Aβ42 peptide–genistein). The average helical percentage of Aβ42 peptide with the presence of EGCG is only a little less than that in the control system, which shows that the effect of EGCG on the α-helix of Aβ42 peptide is not obvious, while genistein can significantly reduce the component ratio of α-helical structures of Aβ42 peptide. Moreover, the average percentages of random coils in the Aβ42 peptide–EGCG system (28.00%) and the Aβ42 peptide–genistein system (29.52%) are markedly increased relative to that in the Aβ42 peptide system (24.69%). The proportions of β-sheet structures greatly decrease from 2.75% in the control system to 0 in systems with EGCG and genistein. Taken together, EGCG and genistein can enhance the ratios of random coil structures and especially reduce the content of β-sheet secondary structures. The results indicate that EGCG and genistein can cause structural disorder to Aβ42 peptide and have effectively impeded the conformational transitions of Aβ42 peptide.

Hydrogen Bond Analysis

In order to further investigate the mechanisms of inhibiting conformational changes of Aβ42 peptides by EGCG and genistein, we analyze the effects of two polyphenolic molecules on the hydrogen bonds (H-bonds) of Aβ42 peptides. As shown in Figure , the average number of H-bonds has stabilized at about 27 ± 3 (Aβ42 peptide), 26 ± 2 (Aβ42 peptide–EGCG), and 25 ± 4 (Aβ42 peptide–genistein). Compared with the H-bond number of Aβ42 peptide in the control system, the H-bond break rates of Aβ42 peptide are 3.70% (Aβ42 peptide–EGCG) and 7.41% (Aβ42 peptide–genistein). In other words, the average number of H-bonds has slightly decreased with the existence of EGCG and genistein, respectively, which implies that the intramolecular H-bonds of the peptide chain partially dissociate. Previous studies have shown that the intramolecular H-bonds of Aβ peptide play a critical role in structural stability and amyloid peptide aggregation.[32] Thus, EGCG and genistein show inhibition against conformational changes of Aβ42 peptides.

Figure 9

Comparison of the H-bond numbers in Aβ42 peptide.

Binding Free Energy Analysis

To further confirm the intermolecular interactions, we have calculated the binding free energy (ΔGbinding) of Aβ42 peptide with respect to EGCG and genistein. The last 30 ns trajectories with Δt = 100 ps of the Aβ42 peptide–EGCG system and the Aβ42 peptide–genistein system are collected and calculated by the MM-PBSA method. Various energy terms are depicted in Table . The average value of ΔGbinding between the Aβ42 peptide and EGCG is about −20.06 ± 4.62 kcal/mol, wherein the molecular mechanics potential energy in a vacuum (ΔEMM = −47.06 ± 8.11 kcal/mol), which is favorable for binding, consists of the van der Waals interactions (ΔEvdW = −32.38 ± 8.14 kcal/mol) and electrostatic interactions (ΔEelec = −14.68 ± 0.21 kcal/mol), while the sum of the polar contribution (ΔGps = 30.90 ± 6.03 kcal/mol) and the nonpolar contribution (ΔGnps = −3.90 ± 0.72 kcal/mol) is the Gibbs free energy of solvation (ΔGsolv = 27.00 ± 5.37 kcal/mol), which is unfavorable for binding. In summary, the van der Waals interactions between the Aβ42 peptide and EGCG play an important part in the binding affinities. In other words, the hydrophobic interactions are dominant in the intermolecular energy of Aβ42 peptide with EGCG. The results are also consistent with previous studies, which have reported that the nonpolar interactions contribute more than 71% to the binding free energy of the EGCG–Aβ42 peptide complex.[31]

Table 2

Various Energy Terms of the Binding Free Energy of Aβ42 Peptide with Respect to EGCG and Genisteina

energy terms	Aβ42 peptide–EGCG (kcal/mol)	Aβ42 peptide–genistein (kcal/mol)
ΔEvdW	–32.38 ± 8.14	–9.01 ± 3.80
ΔEelec	–14.68 ± 0.21	–41.58 ± 7.53
ΔEMMb	–47.06 ± 8.11	–50.59 ± 7.90
ΔGps	30.90 ± 6.03	49.59 ± 8.54
ΔGnps	–3.90 ± 0.72	–2.59 ± 0.27
ΔGsolvc	27.00 ± 5.37	47.00 ± 8.32
ΔGbindingd	–20.06 ± 4.62	–3.59 ± 0.78

The unit of each energy term is kcal/mol.

ΔEMM = ΔEvdW + ΔEelec.

ΔGsolv = ΔGps + ΔGnps.

ΔGbinding = ΔEMM + ΔGsolv.

However, the average value of ΔGbinding (−3.59 ± 0.78 kcal/mol) is relatively small for the Aβ42 peptide–genistein system. The electrostatic term (ΔEelec = −41.58 ± 7.53 kcal/mol) makes an excellent contribution to the binding free energy. While the average value of the van der Waals interactions (ΔEvdW = −9.01 ± 3.80 kcal/mol) is small, the average value of ΔGnps is −2.59 ± 0.27 kcal/mol. Compared to other energy terms, the polar term (ΔGps = 49.59 ± 8.54 kcal/mol) is more unfavorable for binding. Therefore, the results suggest that genistein has a weaker interaction with Aβ42 peptide than EGCG.

In addition, we have investigated the binding free energy contribution of each residue of Aβ42 peptide in the Aβ42 peptide–EGCG system and the Aβ42 peptide–genistein system. Previous studies have shown that the residues, which contribute less than −1.0 kcal/mol to the binding free energy, are defined as the important residues for binding to ligand.[33]Figure a shows the contribution of each residue of Aβ42 peptide to the binding free energy in the Aβ42 peptide–EGCG system. The residue Phe19 contributes the largest energy to the binding of Aβ42 peptide with EGCG. The reason for this could be that the residue Phe19 in the central hydrophobic core has strong hydrophobic interactions with the EGCG molecule, which is in agreement with the reported result.[34] As shown in Figure b, the binding free energy contribution per residue of Aβ42 peptide in the Aβ42 peptide–genistein system is above −1.0 kcal/mol, which means that the binding affinity between genistein and Aβ42 peptide is low. In sum, the results of binding free energy analysis demonstrate that EGCG can indeed effectively inhibit the conformational transition of Aβ42 peptide, while genistein shows less ability to prevent the conformational change of Aβ42 peptide.

Figure 10

Contribution of each residue of Aβ42 peptide to the binding free energy in (a) the Aβ42 peptide–EGCG system and (b) the Aβ42 peptide–genistein system.

Conclusion

The misfolded conformation of Aβ42 peptide plays a dominant role in the initial formation of Aβ plaques. In this work, we have investigated the effects of EGCG and genistein on the conformational evolution of Aβ42 peptide by using molecular docking and MD simulation. The two specific binding regions of Aβ42 peptide have been obtained by the DoGSiteScorer tool. The results of the molecular docking simulation show that EGCG and genistein have the optimal binding conformations with binding region 2 of Aβ42 peptide. In addition, MD simulation studies illustrate that EGCG and genistein can block the conformational transitions of Aβ42 peptides. Compared with the inhibitory effect of genistein, EGCG has stronger intermolecular interactions with Aβ42 peptide. The results have also been demonstrated by the MM-PBSA method, indicating that the hydrophobic interactions between EGCG and the central hydrophobic core of Aβ42 peptide play an important role in the affinities. Therefore, EGCG is very promising in inhibiting the misfolded conformation and self-assembly of Aβ42 peptide.

Methods

Structure Preparation

The three-dimensional (3D) structure of Aβ42 monomer (PDB ID: 1IYT) is obtained from the protein databank (PDB). The sequence of Aβ42 peptide is D1AEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA42. 1IYT is the 3D NMR structure of Aβ42 peptide, which is determined as the full-sized structure comprising residues 1–42 stably and the length and position of two helical regions accurately in the medium of aqueous solutions of fluorinated alcohols.[25] Therefore, 1IYT as a topology model is applied to study conformational changes with or without inhibitors.[11] In addition, the topologies of Aβ42 monomer populate two distinct states: one is free of secondary structure, and the other has some helical regions, which are determined by the deep learning AlphaFold2 method, and can also be used to help design new inhibitors.[35] Model 1 of the NMR conformation in PDB file 1IYT is used for docking and MD simulation. AutoDockTools 1.5.6 software[36] is used for the conversion of PDB file formats. The structures of the EGCG molecule (CID: 65064) and genistein molecule (CID: 5280961) are taken from PubChem. The optimizations of the two polyphenols are performed with the Hartree–Fock method with the 6-31G(d) basis set by using Gaussian09W, Revision A.02, software.[37] EGCG and genistein are submitted to Automated Topology Builder and Repository version 3.0 (ATB3.0)[38] to get the GROMOS96 force field parameters for MD simulation.

Binding Region Prediction

The specific binding sites of Aβ42 peptide with natural polyphenolic compounds have not been reported. Therefore, the binding regions of Aβ42 peptide are predicted by the DoGSiteScorer tool of ProteinsPlus, which is an online server. DoGSiteScorer can be employed to predict binding sites and estimate druggability.[39] A visualization of the binding sites is performed with UCSF Chimera.[40] Subsequently, binding regions are obtained and used for the following molecular docking simulation.

Molecular Docking Simulation

The AutoDock 4.2 software[41] can be utilized to perform the molecular docking simulations between Aβ42 peptide and the two polyphenolic molecules. AutoDock 4.2 is a widely used docking program with exceptional accuracy.[42] The gasteiger charges and polar hydrogens are added to Aβ42 peptide and the two polyphenolic molecules, respectively. Aβ42 peptide is selected as a rigid receptor, and small molecules are set as flexible ligands throughout the docking process. The grid box for binding region 1 of Aβ42 monomer is set to 54 Å × 40 Å × 60 Å with the grid center defined as x = 3.741, y = −1.434, and z = −5.781. The grid spacing is 0.375 Å. The Lamarckian genetic algorithm (LGA) is used for the stochastic search algorithm of docking.[43] Specifically, the binding conformations for ligands are generated with the maximum number of energy evaluations (2.5 × 106). All other parameters are left at the default settings.

For the docking between binding region 2 of Aβ42 peptide and the two polyphenolic molecules, the setting of each parameter is quite similar to that of binding region 1, except that the grid box is set to 42 Å × 42 Å × 44 Å with the grid center defined as x = −3.775, y = −2.583, and z = 7.965. All other arguments are implemented in the same way as those of binding region 1. AutoDockTools 1.5.6 and visual molecular dynamics (VMD) software[44] are utilized for the analysis and visualization of the results of the docking between the two binding pockets and small molecules. In order to do further study on the interaction mechanisms of the optimal active pockets of Aβ42 peptide with the favorable docking conformations of EGCG and genistein, molecular dynamics simulations are performed.

Molecular Dynamics Simulation

GROMACS package 5.1.4[45] is applied to all MD simulations. First, the models of the Aβ42 peptide–EGCG complex and the Aβ42 peptide–genistein complex are constructed by the Aβ42 peptides and the optimal binding poses of EGCG and genistein. The model of Aβ42 peptide in water is used as the control system. The GROMOS96 54a7 force field,[46] which has reported that the simulation results of Aβ peptide are consistent with experimental data,[47] is chosen to model the potential parameters of Aβ42 peptide. The protonation states of the N-terminus and C-terminus of Aβ42 peptide are assigned according to the physiological pH. Periodic boundary conditions defined in all directions are performed in the 7.28 × 7.28 × 7.28 nm3 cubic box. The SPC water model[48] is chosen as the water solvent, and three sodium ions as counterions are added to neutralize the negative charges on Aβ42 peptide. Specific parameters are listed in Table . Second, energy minimization of each system is performed by the steepest descent algorithm. The NVT and NPT ensembles[49] are adopted to equilibrate the system before measurement. The temperature 300 K and 1 bar pressure are determined by the V-rescale thermostat[50] and Parrinello–Rahman barostat isotropically.[51] The LINCS algorithm[52] is utilized to constrain the bonds of Aβ42 peptide and the two polyphenolic molecules. The cutoff value of short-range van der Waals interactions is set to 14 Å, and the long-range electrostatic interactions are calculated by the particle mesh Ewald (PME) method.[53]

Table 3

Specific Parameter Settings of Three System Models

system model	simulation time (ns)	number of total atoms in the simulation box	number of counterions (Na+) in the simulation box
Aβ42 peptide	100 × 3a	37234	3
Aβ42 peptide–EGCG	100 × 3a	37258	3
Aβ42 peptide–genistein	100 × 3a	37246	3

Measurements have been performed three times for each model using different initial velocities.

In addition, GROMACS provides extremely high performance compared to other programs.[54] Since the three systems can fully attain the first metastable states around the starting structures within 100 ns MD simulations on the basis of a plateau in the RMSD profile, MD simulations for the systems are monitored for 100 ns with a time step of 2 fs.[55] To validate the reproducibility and statistical significance of the results, measurements are performed three times for the three models using different initial velocities, and the data are expressed as average values. The MD trajectories are analyzed and visualized by the tools of GROMACS and VMD. The RMSD is subject to the peptide backbone atoms relative to their initial conformations. The secondary structural data of Aβ42 peptide is calculated by the STRIDE algorithm,[56] and the secondary structural contents are identified by the definition of secondary structure of proteins (DSSP).[57] The binding free energy and energy contribution per amino acid residue of Aβ42 peptide in the systems with EGCG and genistein are calculated by the g_mmpbsa package,[58] which uses the molecular mechanism Poisson–Boltzmann surface area (MM-PBSA) method[59] for GROMACS.